Multi-objective Optimization for Common and Special Components: First Step Toward Network Optimization of Regular and Non-Regular Flights
Multi-objective Optimization for Common and Special Components: First Step Toward Network Optimization of Regular and Non-Regular Flights
复制标题
DOI：
10.1142/s1793005715400050
复制
发表时间：
2015-05
期刊：
影响因子：
--
通讯作者：
Takahiro Jinba;Hiroto Kitagawa;Eriko Azuma;Keiji Sato;Hiroyuki Sato;K. Hattori;K. Takadama
中科院分区：
文献类型：
--
作者：
Takahiro Jinba;Hiroto Kitagawa;Eriko Azuma;Keiji Sato;Hiroyuki Sato;K. Hattori;K. Takadama
为了优化由（i）应从所有目标函数的角度优化的公共组件和（ii）应从其中一个目标函数的角度优化的特殊组件组成的问题，本文提出了一种新的多目标优化方法，该方法不仅优化所有目标函数的公共组件，而且还优化每个目标函数的特殊组件。为了研究所提出方法的有效性，本文在测试台问题上测试了我们的方法，该问题是 0/1 背包问题的扩展版本。密集的实验揭示了以下含义：（i）我们的方法找到了比传统方法（NSGA-II）具有更高适应度的更好的解决方案； (ii) 我们的方法可以找到具有大范数（对应于航空公司在航班调度问题中的高利润）且公共成分率高的解决方案； (iii)由于我们的方法中采用的拥挤距离有助于在解搜索过程中保持多样性，因此我们的方法具有很高的解探索能力。
To optimize the problem composed of (i) the common components which should be optimized from the viewpoint of all objective functions and (ii) the special components which should be optimized from the viewpoint of one of the objective functions, this paper proposes a new multi-objective optimization method which optimizes not only the common components for all objective functions but also the special ones for each objective function. To investigate the effectiveness of the proposed method, this paper tested our method on the test-bed problem which is an extended version of the 0/1 knapsack problem. The intensive experiments have revealed the following implications: (i) Our method finds better solutions which have higher fitness than the conventional method (NSGA-II); (ii) our method can find the solutions that had a large norm (which corresponds to a high profit of an airline company in the flight scheduling problem) with the high rate of the common components; and (iii) since the crowding distance employed in our method contributes to keeping the diversity during the solution search, our method has high exploration capability of solutions.